Discovery and structure of a widespread bacterial ABC transporter specific for ergothioneine

Abstract

L-Ergothioneine (ET), the 2-thioimidazole derivative of trimethylhistidine, is biosynthesized by select fungi and bacteria, notably Mycobacterium tuberculosis, and functions as a scavenger of reactive oxygen species. The extent to which ET broadly functions in bacterial cells unable to synthesize it is unknown. Here we show that spd_1642-1643 in Streptococcus pneumoniae, a Gram-positive respiratory pathogen, encodes an ET uptake ATP-binding cassette (ABC) transporter, designated EgtU. The solute binding domain (SBD) of EgtU, EgtUC, binds ET with high affinity and exquisite specificity in a cleft between the two subdomains, with cation-π interactions engaging the betaine moiety and a network of water molecules that surround the thioimidazole ring. EgtU is highly conserved among known quaternary amine compound-specific transporters and widely distributed in Firmicutes, including the human pathogens Listeria monocytogenes, as BilEB, Enterococcus faecalis and Staphylococcus aureus. ET increases the chemical diversity of the low molecular weight thiol pool in Gram-positive human pathogens and may contribute to antioxidant defenses in the infected host.

Fig. 1. spd_1643-1642 encodes an ET uptake transporter in S. pneumoniae denoted EgtU.

a Hypothesized functional roles of the genes of the streptococcal conserved operon harboring a candidate QAC ABC transporter, encoded by spd_1642-1643. b Structure of HPE-IAM-derivatized (capped) ET and LC traces of capped ET from authentic ET (standard) and cell lysates from the indicated S. pneumoniae D39 strains grown on BHI media. WT, wild-type; ∆egtU, markerless deletion of most of the egtU gene, which was then repaired via insertion of a wild-type egtU allele (∆egtU repaired). c MS1 (left) and LC-MS/MS (right) spectra of HPE-IAM capped ET found in a wild-type (WT) cell lysate from cells grown in BHI vs. authentic ET (standard). d Normalized content of ET, GSH and CYS found in the indicated strains of S. pneumoniae D39 grown in BHI in biological triplicate. *, not detected (≤0.0001 nmol thiol/mg protein). e Normalized content of ET and CYS in the indicated strains of S. pneumoniae D39 grown in CDM supplemented with indicated concentration of ET in biological triplicate. *, not detected (≤0.001 nmol/mg protein). GSH is not detected in these cell lysates (≤0.0001 nmol/mg protein). Molar thiol concentrations estimated from nmol/mg protein in S. pneumoniae as described (see also Supplementary Fig. 5). f Same as d but with the indicated egtUA mutant strains. g ATPase activity of WT vs. mutant EgtUA proteins in vitro. **p < 0.01, ***p < 0.001 in a one-sided t-test. Data for d, e, f, and g are shown as the mean and standard deviation of three independent replicates, with individual measurements shown as red circles. Source data for d, e, f and g are provided as a Source Data file.

Fig. 2. Structure of ET-bound SpEgtUC.

a Titration of ET into 1.0 µM SpEgtUC monitored by change in intrinsic Tyr fluorescence. Each titration point is shown as the mean and standard deviation of three independent replicates. Inset, tyrosine emission spectra of SpEgtUC in the absence (black) and presence (red) of saturating ET. b Crystal structure of ET-bound EgtUC_CTT shown as ribbon, with D1 shaded light blue (residues 232–331) and dark blue (445-503), D2 shaded gray (341-432) and linkers colored red. ET is shown as cyan sticks. c Electron density map of ET and surrounding residues in the ligand binding pocket. d Quaternary amine region of the ET binding pocket, with residues in the aromatic pentagon shown as sticks, with polar and cation–π interactions shown as yellow dashed lines, with distances shown in Å. e ET binding pocket of D2, with backbone ribbon colored as in a. Water molecules within 4 Å of heavy atoms are shown as red spheres. Side chains in contact with water molecules or ET are shown as sticks. Polar interactions less than 4 Å are shown as yellow dashed lines. f The D1 ET binding pocket displayed as in d, with C-H•••S hydrogen bonds shown as gray dashed lines (Supplementary Fig. 9c). g High-occupancy water molecules (a subset are labeled 1-9 in panels d and e; see Supplementary Table 3) lining the binding pocket and interdomain cleft, shown as red spheres. h Overlay of ligand binding pockets from the GB SBP AfProX (magenta, PDB 1SW2) and SpEgtUC (blue/red/cyan), showing that a conserved G244 in SpEgtU SBD provides space for a chain of water molecules (red spheres). The F-to-Y switch between EgtU and AfProX homologs is also labeled, with F293 and Y337 in SpEgtUC and Y63 and F107 AfProX shown as sticks.

Fig. 3. Thermodynamics of ET binding to wild-type and mutant SpEgtUCs.

a–e Thermograms and representative ITC-derived ET binding curves obtained for WT and indicated mutant EgtUCs. The continuous lines show the best fit to a single-site binding model. Thermodynamic parameters are compiled in Table 1. Each titration shown is representative of at least two independent replicates. Source data are provided as a Source Data file.

Fig. 4. Conformational and dynamic changes in SpEgtUC upon ET binding.

a AlphaFold2 model of “open” apo SpEgtUC (gray D2, green D1, magenta linkers). F277W and the L374C site of bimane labeling are shown as an open hexagon and a cyan circle, respectively. b The “closed” crystal structure of ET-bound SBD (gray D2, blue D1, red linkers), with F277W and L374C shown as in a. c Bimane fluorescence emission spectra of bimane-labeled L374C/F277W SpEgtUC with (red) and without (black) ET. d Normalized change in fluorescence emission at 480 nm upon titration of bimane-labeled L374C/F277W SpEgtUC with ET in a position-induced fluorescence quenching (PiFQ) experiment. Each titration point is shown as the mean and standard deviation of two independent replicates. See Table 1 for fitted parameters. e ¹H,¹⁵N TROSY spectrum of apo EgtUC (residue-specific assignments in Supplementary Fig. 12). f ¹H,¹⁵N TROSY spectrum of EgtUC bound to equimolar ET (Supplementary Fig. 13). g Backbone chemical shift perturbations (CSPs) upon binding ET for each residue in SpEgtUC. Assignments are missing for residues H310 and V485 in the ET-bound state (shaded pink). Prolines are shaded gray. h CSPs of ET binding painted onto the crystal structure of ET-bound SpEgtUC, with large chemical shift changes shown as thick, red tubes. i B-factors plotted on the crystal structure of SpEgtUC_CTT, with high values shown as thick, red tubes, revealing low B-factors in the interdomain linkers, and comparable to the B-factors of high occupancy solvent molecules. j Overlay of ¹H,¹⁵N TROSY spectra of apo and ET-bound EgtUC, zoomed to the region where the arginine side chain peaks are folded into this spectral window. k H-bond networks of arginine side chains (R320, R379, R404) found in the crystal structure. l ¹H,¹⁵N heteronuclear NOEs for these three arginine side chains. Heteronuclear NOE data were recorded as one replicate, with error bars indicating the uncertainty derived from spectral noise. Source data for those data shown in panel d are provided as a Source Data file.

Fig. 5. The SpEgtUC binds ET with high selectivity.

a Upper, schematic representation of the EgtUC-GFP fusion protein construct; lower, ribbon representation of an AlphaFold2 model of the EgtUC-GFP fusion, with domains indicated. b GFP fluorescence emission spectra, λ_ex = 470 nm with or without 10 µM ET. c ET binding to the EgtUC-GFP fusion protein, monitored by quenching of the GFP fluorescence. Each data point is shown as the mean and standard deviation of three independent replicates. Continuous curve, fit to a 1:1 binding model; see Table 1 for binding parameters. d same as c except L-hercynine was added to EgtUC-GFP fusion protein (Table 1). e Quenching of GFP fluorescence of the EgtUC-GFP fusion protein following addition of 1, 10, or 100 µM of the indicated ligand (see Supplementary Fig. 14a for chemical structures). HIS, L-histidine; PB, proline-betaine; CHO, choline; ECT, ectoine; CAR, carnitine; DMSP, dimethylpropiothetin. Each bar represents triplicate measurements with each data point represented by a filled circle. *p < 0.05 in a one-sided t-test. f Same as e, except that 1 µM ET (left bar) was compared to a mixture of 1 µM ET and 100 µM of the indicated ligand (other bars). Each bar represents triplicate measurements (filled circles). g and h Movement of the indicated backbone NH crosspeak from the apo-state (black) as L-hercynine (HER) is added (yellow to purple), compared to the cross peak position of ET-bound EgtUC (red). i Backbone chemical shift perturbation (CSP) maps resulting from the addition of 2 mM HER or 30 mM glycine-betaine (GB). j Backbone CSP maps upon HER binding painted onto the SpEgtUC structure. k Global fits of the movement of selected NH cross-peaks as a function of [HER] to a 1:1 binding model are shown (Table 1). l Backbone CSPs upon HER binding painted onto the SpEgtUC structure. Source data for c–f, i, and k are provided as a Source Data file.

Fig. 6. EgtU homologs cluster in a grouping of highly similar sequences within a subcluster of SSN cluster 2.

a Several EgtU sequence metanodes are highlighted with a large yellow circle, including L. monocytogenes BilEB and E. faecalis EgtUC biochemically characterized here. The lower (SpEgtU-like), middle, left and upper subclusters of SSN cluster 2 are labeled, with full sequence conservation maps shown in Supplementary Fig. 19. The middle subcluster sequences are derived from anaerobes or obligate anaerobes, including those recently studied in C. difficile. b Representation of SSN cluster 2 colored according to neighborhood connectivity (NC; see scale), with those sequences within a single metanode that are most closely related characterized by a large NC index (and shaded magenta). c Sequence logo representation of conservation in the SBD of EgtUs of those closely related sequences encircled by the red box in b. Secondary structure of the SBD is indicated, based on the structure of SpEgtUC. Residues discussed are highlighted with a specific symbol of interaction, as indicated. See text for additional details. TMD, predicted sites of interaction with the transmembrane domain. Source data for these images shown in panels a, b are provided as a Source Data file.